Summary Report of My Scientific Activities during ERCIM Postdoc fellowship at NTNU

Motif discovery is a crucial part of regulatory network identification, and therefore widely studied in the literature. Motif discovery programs search for statistically significant, well-conserved and over-represented patterns in given promoter sequences. When gene expression data is available, there are mainly three paradigms for motif discovery; clusterfirst, regression, and joint probabilistic. The success of motif discovery depends highly on the homogeneity of input sequences, regardless of paradigm employed. In this work, we propose a methodology for getting homogenous subsets from input sequences for increased motif discovery performance. It is a unification of cluster-first and regression paradigms based on iterative cluster re-assignment. The experimental results show the effectiveness of the methodology. 5 Status: Accepted for publication in the Proceedings of Computational Systems Bioinformatics (CSB 2006) conference, August 14-18, 2006, Stanford University, California. 4.2 Paper 2 Title: TScan: A Two-step De novo Motif Discovery Method. Authors: Osman Abul, Geir Kjetil Sandve, and Finn Drabløs Abstract: Computational discovery of novel motifs in biological sequences is an important and well-studied problem. The key to motif discovery methods, either de novo or library based, is having well-defined scoring functions. Several different scalar valued scoring functions have been proposed that measure some notion of biological motifs; that is we lack a perfect one capable of measuring of all notions together. In this work, we propose a two-step de novo motif discovery paradigm employing two scoring functions measuring different notions of biological relevance. We define a word-counting based method, called TScan, taking this paradigm. It is mainly inspired from MDScan, but does not require supplementary ChIP-chip data. Our results on seven data sets from a recent study are promising, with discovered motifs agreeing well with the consensus motifs defined for the data sets. Computational discovery of novel motifs in biological sequences is an important and well-studied problem. The key to motif discovery methods, either de novo or library based, is having well-defined scoring functions. Several different scalar valued scoring functions have been proposed that measure some notion of biological motifs; that is we lack a perfect one capable of measuring of all notions together. In this work, we propose a two-step de novo motif discovery paradigm employing two scoring functions measuring different notions of biological relevance. We define a word-counting based method, called TScan, taking this paradigm. It is mainly inspired from MDScan, but does not require supplementary ChIP-chip data. Our results on seven data sets from a recent study are promising, with discovered motifs agreeing well with the consensus motifs defined for the data sets. Status: Submitted for publication to The Third Annual RECOMB Satellite Workshop on Regulatory Genomics conference, July 17-18, 2006, National University of Singapore, Singapore. 4.3 Paper 3 Title: Bias Analysis of Motif Models for Biosequences. Authors: Geir Kjetil Sandve, Osman Abul, Vegard Walseng, and Finn Drabløs Abstract: The discovery of motifs in biosequences is an important problem and has in recent years attracted much research interest, resulting in more than hundred tools. The The discovery of motifs in biosequences is an important problem and has in recent years attracted much research interest, resulting in more than hundred tools. The